STACK: Adversarial Attacks on LLM Safeguard Pipelines
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (44) · 37728-37737
DOI: 10.1609/aaai.v40i44.41108 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Ian R. McKenzie und 7 weiteren Autor:innen.
Abstract (Originalsprache)
Frontier AI developers are relying on layers of safeguards to protect against catastrophic misuse of AI systems. Anthropic guards their latest Claude 4 Opus model using one such defense pipeline, and other frontier developers including Google DeepMind and OpenAI pledge to soon deploy similar defenses. However, the security of such pipelines is unclear, with limited prior work evaluating or attacking these pipelines. We address this gap by developing and red-teaming an open-source defense pipeline. 1 First, we find that a novel few-shot-prompted input and output classifier outperforms state-of-the-art open-weight safeguard model ShieldGemma across three attacks and two datasets, reducing the attack success rate (ASR) to 0% on the catastrophic misuse dataset ClearHarm. Second, we introduce a STaged AttaCK (STACK) procedure that achieves 71% ASR on ClearHarm in a black-box attack against the few-shot-prompted classifier pipeline. Finally, we also evaluate STACK in a transfer setting, achieving 33% ASR, providing initial evidence that it is feasible to design attacks with no access to the target pipeline. We conclude by suggesting specific mitigations that developers could use to thwart staged attacks.
Zitieren
APA
McKenzie, I. R., Hollinsworth, O. J., Tseng, T., Davies, X., Casper, S., Tucker, A. D., Kirk, R., & Gleave, A. (2026). STACK: Adversarial Attacks on LLM Safeguard Pipelines. Proceedings of the AAAI Conference on Artificial Intelligence, 40(44), 37728-37737. https://doi.org/10.1609/aaai.v40i44.41108
Vancouver
McKenzie IR, Hollinsworth OJ, Tseng T, Davies X, Casper S, Tucker AD, et al. STACK: Adversarial Attacks on LLM Safeguard Pipelines. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(44):37728-37737. doi:10.1609/aaai.v40i44.41108
BibTeX
@article{McKenzie2026,
author = {McKenzie, Ian R. and Hollinsworth, Oskar John and Tseng, Tom and Davies, Xander and Casper, Stephen and Tucker, Aaron David and Kirk, Robert and Gleave, Adam},
title = {STACK: Adversarial Attacks on LLM Safeguard Pipelines},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {44},
pages = {37728--37737},
doi = {10.1609/aaai.v40i44.41108}
}
EUVIMED